Sharp dimensional analysis of midpoint methods for Langevin sampling

DOI10.48550/ARXIV.2610.06308MaRDI QIDQ7367842arXivFDO

Authors Sinho Chewi, Jianfeng Lu, Matthew S. Zhang, Fan Chen

Publication date 5 October 2026

Published in arXiv

Tldr We derive sharp bounds for the number of gradient queries needed by midpoint methods for Langevin dynamics and exploit cancellations to improve previous results using Malliavin calculus; our bounds show surprising superiority of deterministic methods over randomized ones in high dimensions.

Full work available at URL https://doi.org/10.48550/ARXIV.2610.06308

arXiv classification

cs.DScs.LGmath.NAmath.ST

This page was built for publication: Sharp dimensional analysis of midpoint methods for Langevin sampling

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7367842)